When using push and pop inside a routine (more than at start and end) I usually keep from making my feeble mind getting me in trouble by doing the following (a tab in my code shown as ............):

mov dptr, #stringpush dpl
push dph ;...................~s1
call display_main_routine
push dpl ;...................~s2
push dph ;..................~s3
call display_subroutine
push dpl ;...................~s4
push dph ;..................~s5
mov dptr, #foo
;execute suboroutine
pop dpl ;....................~s5
pop dph ;....................~s4
pop dpl ;....................~s3
pop dph ;....................`s2
pop dpl ;....................~s1
pop dph ;....................~oops

the tilde (~) allow global searches, the numbering show what registers must match.

In this example you clearly see the error at 5 and 4, 2 and 3, 1 and oops

Erik